DESCRIPTION: (Applicant's Abstract) The goal of the proposed research is to develop accurate algorithms for annotating the human genome by combining the rigor of probabilistic modeling with the power of Comparative genomics. The approach will be to use orthologous human and murine genomic sequences within a probabilistic framework that simultaneously models both the structure of genes (eg. exons, introns, untranslated regions, etc.), and the expected conservation in the each of the components. Supporting algorithms that enable the use of draft sequence and whole-genome shotgun sequences are also proposed in order for researchers to immediately take advantage of this expected improvement in gene prediction technology. The research will be conducted by Dr. Ian Korf, under the supervision of Dr. Warren Gish at the Genome Sequencing Center, in the Department of Genetics, and Dr. Michael Brent, in the Department of Computer Science, at Washington University, St. Louis. From his mentors, Dr. Korf, a molecular biologist by training, will learn the techniques of probabilistic modeling and improve his computer programming skills, in order to attain the stated goals of the project, and to lay a foundation for his career as an independent researcher in the field of computational molecular biology.